Parrot 2.11.0 Released | parrot.org | Log: irclog.perlgeek.de/parrot/today | More gci tasks | Merge html_cleanup | Get Plumage working | Review current embedding api | Study EventHandler | Run examples_tests, test HLLs before merges
Set by moderator on 31 December 2010.
rfw nwellnhof: i can't get the error_output_is to fully conform to the error_output tests 00:00
nwellnhof bacek: i thought more about constant PObjs, and +1 for removing constant PMCs. but i'd like to keep constant strings.
rfw because it has a realy weird way of running commands
and i can't easily get the exit code from that
00:01 Kristaba left
bacek nwellnhof, no way. We can use "constant" flag as hint for GC to allocate such PObj from "very-old-generation". But contract between GC and Parrot must not have such assumption that "constant" objects are never removed. 00:02
nwellnhof bacek: yes, it should be a hint. but for strings this optimization can be implemented very cheaply. strings don't reference other objects, so we don't run into problems like with constant PMCs. 00:05
00:05 bacek left
rfw cotto_work: okay, i wrote a naughty regex to compensate for weird stdin handling 00:10
dalek TT #1028 closed by whiteknight++: can't iterate a dynlexpad 00:11
TT #1028: trac.parrot.org/parrot/ticket/1028
TT #731 closed by whiteknight++: can't tell if an OS file is executable from parrot
TT #731: trac.parrot.org/parrot/ticket/731
cotto_work rfw: ok. Why is that weirdness necessary? 00:12
rfw cotto_work: to be honest i have no idea
mine seems to plonk what i offer from stdin there
but yours disregards that
plobsing nwellnhof: I don't see how that would be an optimization. the performance win in "constant" PMCs is not having to mark them. but strings don't need to be marked anyways. The distinction between "constant" and "very-old, infrequently mark/swept" is meaningless; "constant" as a designation has a lot of baggage. 00:13
cotto_work ok. I'll review it in a few minutes or so.
nwellnhof plobsing: it helps with string buffers. 00:14
plobsing nwellnhof: how? they don't need to be marked.
nwellnhof plobsing: string buffers are compacted. 00:15
plobsing nwellnhof: I don't follow. Are you suggesting that "constant" now also imply "pinned" 00:16
?
nwellnhof plobsing: no, but it simply helps if you don't have to compact the buffers of constant strings.
maybe it's not a big win, but i think it's an easy optimization. 00:17
i'm all for using the constant flag only as a hint to the GC, like bacek said.
00:17 Kovensky left
nwellnhof so it means "the GC *may* keep the object around forever", but it doesn't have to. 00:19
00:19 bacek joined
plobsing so long as we don't have an API for creating mandatory memory leaks like happens now. 00:20
cotto_work rfw: I'd like to know why that regex is necessary. It makes me a bit uneasy. 00:27
it's not a blocker, but it strikes me as something that shouldn't be needed.
00:28 gbacon left, kid51 is now known as kid51_at_dinner
cotto_work there's also a lot of duplication between pir_stdin_output_like and _is 00:29
rfw cotto_work: yeah, i don't know either 00:30
cotto_work you should very easily be able to factor out the common code in those two functions. The tests look fine, so do that and I'll pass. 00:31
rfw should i make a sub _pir_stdin_output_slurp? 00:32
cotto_work whatever fits in with the surrounding code's style
rfw looks like it 00:33
00:35 aantn left
rfw cotto_work: done 00:35
00:36 bacek_ joined
dalek rrot: d28f54d | rfw++ | / (3 files):
Hopefully fixed debug_break test; moved pir_stdin_output_is to Parrot::Test.
 00:40
cotto_work rfw: done and thanks
dalek rrot: 279b451 | rfw++ | lib/Parrot/Test.pm:
Factored out common code in Test.pm.
rrot: 256acfd | cotto++ | / (6 files):
Merge branch 'master' of github.com:parrot/parrot
rfw cotto_work: no, thank you :D
00:41 bacek left 00:42 gbacon joined
nwellnhof i'm going to merge nwellnhof/unicode_io now if noone objects. 00:50
cotto_work nwellnhof: how does it do with Rakudo? 00:51
nwellnhof cotto: i didn't test the latest revision, but it did test it extensivley with Rakudo's spectest a while before. 00:52
cotto_work nwellnhof: ok. That at least implies that there are no major design problems. 00:53
thanks for the fix for the implicit :main problems 00:54
nwellnhof cotto: which one?
cotto_work both, but I was thinking of the revert
nwellnhof cotto: i think someone else made those commits. 00:55
cotto_work indeed. it was plobsing
plobsing++ then
chromatic I still think removing implicit :main makes sense. 00:57
cotto_work chromatic: me too but only when Rakudo and nqp-rx are ready. I like reducing the amount of magic we rely on.
plobsing it won't be much work to remove. it was something like a 10 line patch. 00:58
and it is determined *statically*, so it doesn't exist except in the minds of IMCC users
chromatic Yeah, that's why Patrick's concern confused me. 00:59
nwellnhof src/packfile/api.c:937: warning: ā€˜mainā€™ is usually a function 01:00
whiteknight plobsing: Whe you get a chance can you take a look at the NCITasklist page on the trac wiki, and let me know if it's still active/current?
dalek tracwiki: v4 | whiteknight++ | NamespaceTasklist 01:01
tracwiki: +some more task ideas
tracwiki: trac.parrot.org/parrot/wiki/Namespa...ction=diff
plobsing silly C compiler, trying to tell me what I can and can't name my variables. I have C++ for that! 01:02
whiteknight msg Coke can you take a look at the BuildTasklist page on the wiki and let me know if it's still relevant/active/accurate? Thanks 01:03
aloha OK. I'll deliver the message.
nwellnhof testing 'make moretest' now 01:04
plobsing whiteknight: that tasklist (goallist really) was written back when I thought parrot wanted a "fat" NCI. I am now of the opinion that parrot needs a slim, bare-minimum NCI. 01:06
whiteknight plobsing: Okay, so is it worthwhile to salvage that page, or do we delete it and start over with a new tasklist?
01:08 mikehh left
plobsing whiteknight: it has the completely wrong mindset (esp. when considered in context of lorito). I've scrapped it. 01:08
whiteknight plobsing: okay. I'm doing some wiki cleanups now. You want me to delete it? 01:09
plobsing counted strings, for example, are *not* something that should be happening at the C layer.
whiteknight: note the use of the past tense. it is done.
cotto_work NCI and Lorito is tricky. We don't currently have a good solution apart.
from thunks, and they aren't really that great
plobsing cotto_work: as a rule of thumb though, if it *can* happen parrot-side, it should 01:10
cotto_work "parrot-side"? Parrot has many sides.
chromatic As opposed to C-side, which is Smalltalk. 01:11
dalek rrot: 5690950 | nwellnhof++ | / (149 files):
Merge branch 'master' into nwellnhof/unicode_io
rrot: 7cad06a | nwellnhof++ | src/io/buffer.c:
Fix a compiler warning
rrot: 6f96b02 | nwellnhof++ | / (34 files):
Merge branch 'nwellnhof/unicode_io'
plobsing as opposed to inside the shim layer (native-side)
cotto_work ok
plobsing taken to the extreme, we should even get rid of automatic null-padding of strings in the shim-layer. 01:12
that *can* be done by parrot.
there's way too much stuff in NCI that is merely convenient in stead of necessary
cotto_work: thunks aren't all that bad either. if you know what function signatures you want to call in advance, which *is* possible in most cases, it is possible to provide these. 01:16
dalek tracwiki: v181 | whiteknight++ | WikiStart 01:18
tracwiki: Start rearranging tasklist pages by their relevance and current state. We basically have "current", "in active planning", and "everything else".
tracwiki: trac.parrot.org/parrot/wiki/WikiSta...ction=diff
cotto_work plobsing: sure. It just limits future expansion to functions for which we know signatures and means we need to carry around a big chunk of object code.
01:20 mikehh joined
plobsing cotto_work: but it is possible to create dynext libraries to provide additional thunks. these can be provided by library writters fairly easily. 01:20
cotto_work: parrot can already be configured without these thunks 01:21
cotto_work plobsing: It's workable but not ideal. 01:22
plobsing cotto_work: any dynamic approach is going to be unportable. thunks are the only solution as portable as C.
and the thunks only constitute ~4% of my libparrot.so "big chunk of object code" may be a bit of a mischaracterisation 01:26
01:43 bacek__ joined 01:44 dmalcolm left 01:48 bacek_ left
cotto_work seen notfound 01:48
aloha notfound was last seen in #parrot 5 days 1 hours ago saying "Hurrah".
kapace__ i wonder... 01:50
seen kapace__
aloha kapace__ was last seen in #parrot 0 seconds ago saying "seen kapace__".
kapace__ in my mind, aloha got caught in infinite recursion, and exploded 01:51
cotto_work kapace__: seen kapace__ 01:54
kid51_at_dinner Is NQP building once again? 01:55
01:55 kid51_at_dinner is now known as kid51
cotto_work kid51: it was building before. Some tests look broken. 02:01
(building before, iirc)
whiteknight msg dukeleto can you take a quick look at the GitMigration page on the wiki and see if it is worth keeping? Thanks 02:05
aloha OK. I'll deliver the message.
cotto_work decommutes
whiteknight msg cotto Can you evaluate the L1Recap page on the wiki and see if it contains any information worth keeping, if it can be merged into the Lorito page, or if it can be deleted? Thanks 02:06
aloha OK. I'll deliver the message.
dalek rrot: e1a1d78 | jkeenan++ | lib/Parrot/Test.pm:
[codingstd] Correct POD formatting error.
 02:11
02:12 bluescreen left
whiteknight aloha, coverage? 02:14
aloha whiteknight: coverage is cv.perl6.cz or tapir2.ro.vutbr.cz/cover/cover-results/
02:15 bacek_ joined 02:20 bacek__ left, contingencyplan joined
dukeleto whiteknight: git migration wiki page? 02:24
whiteknight: will see if anything is valuable
whiteknight dukeleto: it's called GitMigration
dalek tracwiki: v182 | whiteknight++ | WikiStart
tracwiki: Move the smolder tasklist in with the rest of the tasklists. Other stuff
tracwiki: trac.parrot.org/parrot/wiki/WikiSta...ction=diff
dukeleto whiteknight: you want to delete the page? or just move it somewherE? 02:25
whiteknight dukeleto: that's what I'm asking you. If the page serves no further purpose, we can delete it
cotto ~~ 02:29
cotto is popular
dukeleto whiteknight: you can delete GitMigration, but i think GitObjections should stay around 02:31
whiteknight: and we need a new wiki page to keep track of Git-related stuff. or perhaps there is another wiki page 02:32
whiteknight ok 02:33
cotto dukeleto, you mean one that says "go read git_workflow, then we can talk"? 02:35
;)
dalek rrot: e2e9c56 | plobsing++ | DEPRECATED.pod:
add dep notice for TT #1931
 02:39
tracwiki: v4 | whiteknight++ | ParrotHackList 02:41
tracwiki: REmoved a hack in PCC
tracwiki: trac.parrot.org/parrot/wiki/ParrotH...ction=diff
tracwiki: v183 | whiteknight++ | WikiStart
tracwiki: trac.parrot.org/parrot/wiki/WikiSta...ction=diff
TT #1931 created by plobsing++: [DEPRECATED] advanced NCI parameter types
TT #1931: trac.parrot.org/parrot/ticket/1931
plobsing whiteknight: TT #1931 represents the slimming down of NCI I envision
dukeleto just ran "make test" for parrot on the gcc compile farm in 50 seconds 02:43
cotto: a place to keep track of what we still need to figure out with git 02:44
Our first smoke report from the GCC Compile Farm: smolder.parrot.org/app/projects/rep...tails/2002
pbc_merge tests have wonkiness. I was using TEST_JOBS=10, could be related 02:45
dalek rrot: f7a044b | Whiteknight++ | src/call/args.c:
move error-message creation for named parameter/argument mismatch into a new function. This doesn't even make a dent in the huge fill_params functions, but it's a start
 02:46
rrot: b863c4d | Whiteknight++ | / (38 files):
Merge branch 'master' of github.com:parrot/parrot
whiteknight plobsing: oh, okay 02:47
02:48 whiteknight left
nopaste "kid51" at 192.168.1.3 pasted "t/tools/pbc_merge.t: New test failure observed on Darwin/PPC." (4 lines) at nopaste.snit.ch/27551 02:53
dukeleto seems like pbc_merge.t fails with TEST_JOBS=1 as well 02:54
kid51: i am seeing pbc_merge.t fail on linux as well
pbc_merge.t failure: smolder.parrot.org/app/projects/rep...tails/2004 02:55
plobsing funny, it doesn't fail here
but I know *exactly* why it is printing that
02:56 bacek_ left
dukeleto plobsing: i just installed a fresh parrot repo on a gcc compile farm machine. everything is fresh 02:56
plobsing dukeleto: mine also spits out those messages, but it doesn't cause a test failure. not sure why.
the messages come from pbc_merge encountering multiple packfiles which specify a :main sub. Implicit :main means that *every* packfile generated by IMCC will have a :main sub. 02:57
so whenever merging any packfile generated by IMCC, this helpful warning is printed
kid51 must sleep 03:04
03:05 bacek_ joined 03:06 kid51 left
dalek rrot: 84a7d28 | plobsing++ | frontend/pbc_merge/main.c:
silence pbc_merge warnings
 03:12
plobsing dukeleto: 84a7d28 should resolve your test failures
03:35 gbacon left, bacek__ joined 03:40 bacek_ left 03:57 fbrito joined
rfw hey fbrito 04:02
04:02 bacek__ left
fbrito rfw: hi :D 04:02
rfw i've been taking all the parrot tasks :( 04:03
fbrito rfw: 250 points?! wow! congratulations! 04:04
69 tasks, heheh
rfw fsagsdhshsdgh goddammit
fbrito rfw: I have to say that I am getting a bit tired of GCI :S. I am working a lot to my father and going through some things in my personal life 04:06
rfw i am too, but there's only 7 more days left!
04:06 fbrito left
rfw no, come back fbrito! 04:06
04:07 fbrito joined
fbrito rfw: yeah, true :P 04:09
rfw where did you go before :(
fbrito Pidgin crashed 04:14
first time I see that happening :D
rfw: or what do you mean by "before"? :P 04:15
rfw ah
the crash
i've just been messing around with javascript and stuff
dukeleto plobsing: i will try again 04:16
fbrito: you need something to work on?
fbrito rfw: oh, you mentioned something about a JS physics engine :D
dukeleto rfw: what are you up to?
fbrito dukeleto: hm, actually I am stuck in a code coverage task 04:17
rfw dukeleto: physics, quadtrees and stuff
i'm surprised how fast 1000 insertions into a javascript quadtree is
dukeleto rfw: sounds fun
fbrito: what you stuck on/
fbrito: ?
fbrito test get_attr_keyed on Object PMC 04:18
dukeleto: I made an example with get_attr_str: pastie.org/1424739 but I don't know how to use the Key in this case
04:20 bacek joined
rfw fbrito: is this for the index == -1 branch? 04:24
04:24 cottoo joined, cotto left, cottoo is now known as cotto
rfw fbrito: ah, you're not doing it correctly i don't think 04:24
get_attr_keyed is a VTABLE function 04:25
so you mean
hold on...
fbrito: pastie.org/1427505 04:26
i think this is what you want
fbrito ahh
rfw i'm not sure about add_attribute 04:27
dukeleto fbrito: i need to go eat dinner 04:28
fbrito rfw: have you try to run that code? :D
dukeleto: ok, no problem
rfw fbrito: not yet
i'm just assuming it works :3
fbrito it gives me "set_string_keyed() not implemented in class '' " on object["color"] = "blue" 04:29
rfw set_string_keyed() not implemented in class ''
i lied :(
hold on 04:30
fbrito default.pmc has a comment on set_attr_keyed 04:31
"Default version of keyed attribute set. Discards the key and does a set by the string name passed in."
discards the key -.-
rfw fbrito: pastie.org/1427511
sorry abou that, i'm a terrible person 04:32
dalek rrot/nwellnhof/read_chars: 5210465 | nwellnhof++ | / (5 files):
[io] Switch read to use characters
rfw oh wait 04:33
you want get_attr_keyed
fbrito ah, true
sorry :D
cotto Wow. fill_params just doesn't end.
rfw no, my bad
fbrito tapir2.ro.vutbr.cz/cover/cover-resu...t-pmc.html 04:34
set_attr_keyed is fully tested
I mean, get_attr_keyed
I mean, set_attr_keyed
ahhaha
rfw get_attr_str is fully covered
as well
fbrito I want to test line 396 04:35
rfw ah
04:35 nwellnhof left
rfw get_attr_keyed? 04:35
fbrito and line 186 and 188 (static get_attrib_index_keyed c function) 04:36
rfw huh, how _do_ you call get_attr_keed 04:37
keyed*
fbrito that's my question
it has 2 params, a key and a string -.-
if I want to look for an attrib by it index position, why should I need a string? 04:38
ah, I think I understood 04:39
rfw exists_keyed_str() not implemented in class 'Integer' is what i'm getting :( 04:40
fbrito ah, nevermind -.-
I was trying something like this: pastie.org/1424739 04:41
rfw trac.parrot.org/parrot/wiki/MakeEveryPMCAnObject
i found an example of it
fbrito I have also found this page by googling for "get_attr_keyed" 04:42
rfw ah
fbrito but the example is for get_attr_str
rfw oh
fbrito I think
rfw ah yeah
it is
damn
dalek rrot/pasm-serializer: 7e145a8 | plobsing++ | / (2 files):
add pasmserializer pmc
rrot/pasm-serializer: e8c84ec | plobsing++ | MANIFEST:
mk_manifest_and_skip
rrot/pasm-serializer: 4abf3e3 | plobsing++ | / (2 files):
regularize spacing slightly
 04:43
rrot/pasm-serializer: 33acf3b | plobsing++ | / (2 files):
eliminate leading space
fbrito rfw: I really can't understand why I have to pass a key and a string as parameters 04:47
if I pass a null key, nothing happens to it
bacek fbrito, which function?
rfw i get null pmc access in get_string() when getting a weird attribute
plobsing fbrito: are you passing key objects to functions? that is magical and probably broken.
rfw aaaaaand now i'm getting segmentation fault 04:48
fbrito bacek I am trying to test get_attr_keyed on Object PMC
04:48 contingencyplan left
fbrito actually it is already tested, but there is an untested exception inside it 04:48
tapir2.ro.vutbr.cz/cover/cover-resu...t-pmc.html
and 2 untested lines on get_attrib_index_keyed static c function (function that get_attr_keyed relies) 04:49
rfw nol888 ripped out share_ro, didn't he 04:50
bacek fbrito, it's like C< $P0 = get_attr $P1, ["Foo";"Bar"], "Baz" >
Key is "namespace", Name is localname 04:51
(Yes, I know it's weird)
fbrito, try to fetch some meaningless attribute 04:52
fbrito ohhh 04:55
hm, some meaningless attribute will raise "Null PMC access in get_string()" 04:56
rfw looking at the code, it might be fq_name = VTABLE_get_string(interp, parent_class); 04:57
but i tend to tell horrible lies about these things so don't listen to me 04:58
kapace__ is there a way to get the pointer of a label? 05:00
erm, address
nvm, got it set_addr 05:02
Continuation's get_string returns a string of info about the current instruction etc, does it make sense to check against a hard coded string (Maybe some values will change?) or check by matching a regex? 05:09
fbrito kapace__: sorry, but I can't answer that :P 05:23
going to bed. good night guy
guys*
kapace__ lol ok fbrito
good night :)
dalek rrot: 27c0799 | plobsing++ | DEPRECATED.pod:
add deprecation notice for TT #1904
 05:24
sorear kapace__: fragile tests are ok as long as they fail not ok
kapace__ checks out how to use nok
05:25 fbrito left
kapace__ wait i think i didn't understand that, do I use nok() somehow, or normal is()? 05:30
05:38 rurban_ joined 05:41 rurban left, rurban_ is now known as rurban
plobsing kapace__: use either. the important thing is, when creating tests likely to work incorrectly on occasion (such as regex tests), try to err on the side of false-failures in stead of false-passes. 05:45
kapace__ ok, but this test checks line numbers, and program counters?
sounds like it could fail at any minor change in the source 05:46
sorear That's OK 05:48
plobsing hair trigger failures are OK. not great, but better than the alternative.
kapace__ ok. 05:49
bacek #1650 looks like good GCI task 05:53
dukeleto bacek: you are back! 05:54
bacek: i hope you are feeling better
bacek dukeleto, yes. Will be released from hospital in couple of days. 05:55
dukeleto bacek: you have a laptop to keep you company?
bacek dukeleto, yes. And 3G internet :)
dukeleto bacek: lots of time to hack on stuff! 05:56
bacek dukeleto, 2 days :) 05:57
#1038 looks like another candidate for GCI
And I just need good candidate for "2 days hackatron" :) 05:58
dukeleto bacek: you thinking that you want to hack on something other than the gc? 06:02
bacek dukeleto, GC isn't "2 days task"
dukeleto bacek: i have a branch that is adding ipv6 support, that might be fun for you to hack on :) 06:03
bacek And I'm slightly tired of it.
dukeleto bacek: yeah, that happens
bacek Hmm. I don't have ipv6 support below my NIC. Idea rejected :)
dukeleto bacek: blarg! 06:04
bacek dukeleto, good try, bad luck :)
kapace__ tunnelbroker!
tunnelbroker.net/
dukeleto bacek: what about merging some branches that haven't seen any love ?
bacek: github.com/parrot/parrot/branches 06:05
bacek: what about getting cardinal to work on parrot master? 06:06
bacek: tadzik got it closer to working recently, but hit some exotic bugs 06:07
bacek dukeleto, interesting idea. 06:08
dukeleto bacek: also, nqp-rx is broke on parrot master now
bacek: the encapsulate-main merge broke it
bacek: pmichaud was not pleased
bacek dukeleto, wasn't it reverted?
dalek p-rx: 89a7459 | cotto++ | src/NQP/ (2 files):
add ...
 06:09
dukeleto bacek: maybe. i have been out for the last few hours
p-rx/nom: 98b0b4f | bacek++ | src/ (3 files):
Remove usage of deprecated concat_s_s op
p-rx/nom: 2a73b89 | cotto++ | src/stage0/ (2 files):
fix deprecated concat_s_s usages
dukeleto bacek: i think cardinal now compiles on parrot master, but fails all tests
bacek dukeleto, let me check 06:11
dalek p-rx: 2f2bc9a | cotto++ | src/NQP/ (2 files):
Revert "add ..."

This reverts commit 89a7459bf15f084f034b4572e3121833f22aa6e3.
rdinal: f1b1628 | bacek++ | src/ (5 files):
concat_s_s is gone
 06:16
bacek dukeleto, bad luck. Looks like cardinal wasn't updated to nqp-rx. 06:25
06:35 bacek_ joined 06:40 bacek left
cotto nom sure loves globals 06:44
06:59 bacek__ joined 07:04 bacek_ left 07:41 theory left
rfw oh, completely forgot i had this os dynpmc task claimed 07:47
cotto lawl 07:50
tadzik bacek__: yes it wasn't, it's still using PGE
07:51 bacek__ is now known as bacek
bacek tadzik, yeah... It will take some time to update it to nqp-rx 07:51
tadzik bacek: making it work was my first step. Now the first thing failing that makes the tests fail, is require giving a Null PMC Access 07:52
I tracked down the line when it occurs, but failed to undersand the source of the problem 07:53
bacek tadzik, it's... May it is good idea. I'm not sure.
tadzik bacek: my second thought was "screw this, nqprx goes first", but got convinved the other way here 07:54
bacek tadzik, I usually use second approach :) 07:55
rfw does parrot have rmdir? 07:57
rfw is trying to do naughty things with cwd 07:58
tadzik bacek: I have an unpushed commit, mind giving me a commit bit or can I nopaste it somewhere? 07:59
bacek tadzik, I'm not metacommiter. 08:00
tadzik bacek: wklej.org/id/449890/ 08:03
rfw is it safe to test for a bad rm by calling "rm"("/")? 08:06
chromatic Not really.
rfw i can't think of any other way to do it
chromatic Imagine some fool runs the test with root privileges.
rfw rm("/") only removes empty directory 08:07
chromatic On an operating system written by fools.
rfw -y+ies
(also i'm running tests at root atm, but that's because i'm very lazy)
(and i don't particularly care about this vm)
chromatic As a rule, assume that anything destructive you do in your tests outside of Parrot itself could do Very Bad Things and test defensively.
rfw yeah 08:08
hm
what else could i do...
i need to pass the S_ISDIR test, but not actually be able to remove the directory
why is / so convenient for this?!
chromatic Create a directory with a file in it. 08:09
rfw oh, of course
my brain isn't working quite right today
thanks chromatic
chromatic You're welcome.
bacek tadzik, applied. 08:10
chromatic, aloha
dalek rdinal: 992ef73 | (Tadeusz Sośnierz)++ | src/ (3 files):
Fixes under Parrot master, it now compiles and the REPL works

Signed-off-by: Vasily Chekalkin <bacek@bacek.com>
bacek tadzik, can you give me patch for parrot/CREDITS so dalek can properly attribute you for karma? :) 08:11
tadzik bacek: I'd like to :) 08:12
bacek tadzik, go for it. Format is trivial :)
tadzik bacek: my entry looks almost like yours, go and see. I don't know why doesn't it work properly 08:13
bacek hmm 08:16
seen sorear
aloha sorear was last seen in #perl6 22 mins 53 seconds ago saying "then ... done".
bacek msg sorear Does dalek handle utf-8 from CREDITS properly?
aloha OK. I'll deliver the message.
tadzik looks like it does
+dalek | cardinal: 992ef73 | (Tadeusz Sośnierz)++
bacek tadzik, it's from RSS feed 08:17
I'm asking about properly loading of CREDITS :) 08:18
aloha, dalek?
aloha bacek: dalek is not picking up on trac wiki changes any more
bacek aloha, forget about dalek
aloha bacek: I don't know anything about about dalek.
bacek aloha, forget dalek
aloha bacek: I forgot about dalek.
tadzik I see 08:20
bacek meh.
github.com/Infinoid/dalek-plugins/...armalog.pm
line 6
7
seen Infinoid 08:21
aloha Infinoid was last seen in #parrot 12 days 3 hours ago joining the channel.
bacek oookey.
cotto rfw, accepted
rfw cotto: thanks
bacek cotto, any ideas who can fix dalek a little bit?
cotto what's broken? 08:24
tadzik afk
bacek link to CREDITS file
mikehh bacek: I think sorear mostly deals with dalek, we need to consider branch merges, and stuff like that
bacek cotto,
github.com/Infinoid/dalek-plugins/...rmalog.pm, line 7
mikehh, thanks for ifo
info
cotto oic 08:25
seen Infinoid
aloha Infinoid was last seen in #parrot 12 days 3 hours ago joining the channel.
mikehh bacek: how you doing, you said you were being released from hospital in a couple of days?
bacek mikehh, I will be. 08:26
For now - relaxing in 500km out of Sydney :)
08:27 chromatic left
mikehh bacek: hopefully not to near the floods 08:28
bacek mikehh, actually, this town was flooded couple of weeks ago. 08:29
Wagga Wagga
www.theaustralian.com.au/news/natio...5966425111 08:30
Few week ago actually :)
dalek rrot: 6be431d | kapace++ | t/pmc/continuation.t:
Added tests for Continuation up to 93%
 08:36
rrot: 32765a4 | cotto++ | t/pmc/continuation.t:
Merge branch 'kapace/gci_test_continuation' of github.com/kapace/parrot into kapace-kapace/gci_test_continuation
rrot: f1c0156 | cotto++ | t/pmc/continuation.t:
[t] make some Continuation tests less brittle
cotto Kapace, task approved 08:37
cotto thought bacek was making silly noises
mikehh got to got out for a while - bbl 08:42
cotto Is it a reasonable gci task to take fill_params in src/call/args.c from 400 lines to 300?
bacek cotto, ouch. It will be scary task. 08:49
rfw puu.sh/EFb what the hell is happening here 08:50
how is that even possible 08:51
cotto bacek, we have some scary students.
I had a bad dream about rfw.
;]
bacek Give than "0-args subs parameters checking" :)
#1033
rfw gets some popcorn to listen to cotto recount of the bad dream
cotto actually it wasn't bad, or a dream, or about rfw 08:52
bacek It will consume about few days to figure out how to do it with minimal damages
rfw aw :(
cotto It might have been a sandwich.
08:53 fperrad joined
bacek actually, many-many tests are failing if I switch on "0-args subs param checking" 08:54
cotto rfw, what's confusing?
rfw cotto: how can that line not be executed
nopaste "bacek" at 192.168.1.3 pasted "enable 0-args subs param checking" (17 lines) at nopaste.snit.ch/27552
rfw and the next line work
dalek rrot/zeroargs_subs_checking: 0ba565b | bacek++ | compilers/imcc/pcc.c:
Enable checking of arguments for 0-args subs
 08:56
rfw also chroot doesn't want to work
bacek cotto, GCI task - make zeroargs_subs_checking to pass all tests :)
cotto bacek, is there a ticket for that 08:57
?
nopaste "bacek" at 192.168.1.3 pasted "Current test result" (56 lines) at nopaste.snit.ch/27553
cotto rfw, where's that code?
bacek cotto, #1033
cotto a tiny snipped isn't too helpful
rfw cotto: src/dynpmc/os.pmc
on those lines
cotto bacek, wheeeeee
bacek cotto, I did quick check. Tests are wrong :) 08:58
And have to be updated :)
In many cases just adding ".param pmc foo" or similar 08:59
cotto that doesn't sound too bad 09:00
bacek Good GCI task from my pov
cotto difficult or medium?
bacek "medium boring"
it's just reviewing many-many failing test cases. 09:01
rfw boy, these tests sure involve crazy backflips
actually cotto that uncovered line is really hard to cover 09:02
i can't figure out how to cover it
because it involves unlinking a file you can't unlink
cotto rfw, I can see why that one is tricky 09:03
rfw there aren't a lot of things i can do 09:04
i can't force ELOOP, ENOENT or ENOTDIR, since parrot checks for that 09:05
and it's perhaps a little dangerous to delete system files 09:08
cotto I'd be disappointed if running make test killed my OS.
rfw okay, this is a very dirty and nasty thing to do 09:09
moritz I'd be disappointed by the OS if a 'make test' could kill it
rfw but i'm trying to delete something from /proc
it covers it, but in a horrible, horrible way 09:10
(and fails on win32, of course)
cotto quite
rfw do you mind if i just ignore it
cotto If there's no sane way to test it, the insane ways are still a bad idea. 09:11
rfw okay, not testing that!
hm, i swear i'm covering chroot 09:14
but apparently it's not covered at all
cotto 1033 task published 09:22
rfw, it'd probably about 30 minutes work for you 09:23
bacek++ for thinking of it
rfw cotto: heh, i think i'll let somebody else do it for a change
cotto your call 09:25
dalek rrot/zeroargs_subs_checking: 61052e5 | bacek++ | t/pmc/pmcproxy.t:
Fix one of test as example.
bacek cotto, it's 1033, 1704, 1705 combo :)
cotto it wouldn't be a bad introductory task
dalek rrot/zeroargs_subs_checking: a62931c | bacek++ | t/compilers/imcc/imcpasm/cfg.t:
Mark :main sub explicitely. Part of #1704
rrot/zeroargs_subs_checking: 1c33c8f | bacek++ | t/compilers/imcc/imcpasm/opt0.t:
Fix tests for handle 0-args param checking behavior.
cotto I'm going to go to bed instead of looking up the rest of those tickets. Good night.
rfw cotto: www.google-melange.com/gci/task/sho...9394564760 done
bacek cotto, it's not "introductory" from my pov. At least it requires some knowledge of parrot guts.
rfw parrot dissection? 09:27
cotto nice test cases 09:28
rfw yes one of them involves ripping the current directory out from underneath parrot 09:29
cotto one of the fails though, and I'm going afk
now
'night
rfw night
i'll look at it in the morning
09:33 Kovensky joined 09:42 rfw left
moritz rakudo on newest parrot complains: 09:43
./perl6: symbol lookup error: ./perl6: undefined symbol: Parrot_gbl_set_config_hash_internal
but I can't find any references to that function in the rakudo sources 09:44
09:46 bacek_ joined 09:51 bacek left
bacek_ msg cotto #1033 is actually "difficult" 09:54
aloha OK. I'll deliver the message.
moritz cleaned everything, error went away 09:57
instead I now get double output
./perl6 -e 'say rand'
=> two random numbers, each followed by \\n
bacek_ moritz, linux/amd64? 10:08
moritz bacek_: yes
bacek_ moritz, let me try to reproduce it on linux/ia32 10:09
moritz parrot is currently... "entertaining" for HLL developers (or maintainers)
bacek_ moritz, how so? what was changed dramatically? 10:10
moritz bacek_: it's just that each time I update to a new parrot, something new has stopped working 10:11
often the old errors are gone, and replaced by some new, entertaining stuff. Like double output.
bacek_ moritz, we are kind of ripping deprecated features. Deprecated long time ago actually. 10:12
moritz bacek_: and have all these features also had replacements long time ago?
10:13 contingencyplan joined
bacek_ moritz, some of them just have to be removed. Some have clean migration strategy. If not - we have to revert "deprecation" commit. Similar to #1704 10:13
moritz as I said - "entertaining" 10:15
bacek_ moritz, -Ofun in action! 10:17
moritz, bug confirmed... 10:27
I suspect it's #1704 related
dalek lek-plugins: 372f8ed | infinoid++ | modules/local/karmalog.pm:
Fix the link to CREDITS.
 10:28
Infinoid sorear, diakopter: I'm not sure where dalek actually runs nowadays, so I can't update its source. 10:30
11:02 bacek_ left, bacek_ joined
moritz if I think a bug is harmful enough that it should be fixed before the next release, should I set the severity to "release"? 11:13
moritz just does it 11:14
dalek TT #1932 created by moritz++: rakudo on parrot RELEASE_2_11_0-656-g27c0799 executes main body twice 11:18
TT #1932: trac.parrot.org/parrot/ticket/1932
11:23 Util joined 11:57 gbacon joined 12:31 bacek_ left
dalek TT #731 reopened by coke++: can't tell if an OS file is executable from parrot 12:41
TT #731: trac.parrot.org/parrot/ticket/731
13:10 lucian_ joined 13:15 lucian left 13:24 Coke joined
Coke ~~ 13:37
13:38 rurban_ joined 13:41 rurban left, rurban_ is now known as rurban 13:44 bluescreen joined 13:46 smash joined
smash hello everyone 13:46
tadzik hello
13:56 lucian_ left
atrodo Morning! 14:05
dalek rrot: e9d9914 | Whiteknight++ | src/call/args.c:
add parrot_hash_destroy calls back where I removed them. I thought they were redundant but on second look I think they are necessary. Other small tweaks
 14:34
14:53 bluescreen left 14:59 Khisanth joined 15:08 bluescreen joined 15:11 whiteknight joined 15:17 hercynium joined
dalek rrot: 9a74ad4 | mikehh++ | src/call/args.c:
fix codetest failure - line length
 15:17
15:36 dmalcolm joined 15:37 PerlPilot joined, PerlPilot left 15:38 PerlPilot joined 15:39 frodwith joined 15:44 Kovensky left, fbrito joined
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#2015) fulltest) at 2_11_0-661-g9a74ad4 - Ubuntu 10.10 i386 (g++-4.5) 15:47
15:55 Kovensky joined
dalek Some 'mikehh' person just gave me a 151282 byte commit packet. They need to be more careful in the future. 15:59
16:00 plobsing left
mikehh ha - in html_cleanup - git merge origin/master, git push 16:00
whiteknight nice 16:01
mikehh: you getting close to a merge with that one?
mikehh whiteknight: still working on generating index/header pages, everything else seems ok 16:05
whiteknight ok 16:06
mikehh whiteknight: a whole bunch of stuff in lib/Parrot/docs will probably go 16:07
whiteknight: been mostly working on master, fixin' codetest failures etc. 16:08
16:11 PerlJam left
dalek rrot/html_cleanup: 275bc30 | mikehh++ | docs/index/tools.json:
remove deleted file - tools/docs/make_html_docs.pl
 16:32
Coke mikehh: thanks for keeping that branch up to date. 16:34
rant: logging in on trac.parrot.org loses your page. 16:35
mikehh Coke: I am trying to generate the index/header pages at the moment, did you have any ideas on that 16:37
Coke: I think I have an handle on it but proceeding slowly
the .joon files have the required info I think, need to push them through Parrot::Docs::HTMLPage.pm 16:40
.json 16:41
at the moment all the page references there are generated 16:42
we might want to include other stuff, but can handle that later 16:43
whiteknight Coke: I HATE THAT PROBLEM 16:49
that's probably the single thing I dislike most about Trac 16:50
Coke what is the equivalent of "svn st --no-ignore" in git?
whiteknight what's svn st?
dalek rrot: da7b6ba | Whiteknight++ | src/string/encoding/shared.c:
fix a library lookup bug on windows from the new unicode changes
 16:52
rrot: 7cc276c | Whiteknight++ | src/call/args.c:
Merge branch 'master' of github.com:parrot/parrot
dukeleto ~~ 16:55
Coke whiteknight: "status"
dukeleto Coke: git status
Coke dukeleto: what is the equivalent of "svn st --no-ignore" in git?
dukeleto Coke: answered before you asked ;)
Coke emphasis on "--no-igore"
*ignore
dukeleto Coke: git status -u will also show untracked. Without -u, no untracked are shown
Coke not untracked. ignored. 16:56
let's ignore svn then.
dukeleto Coke: git help status
Coke How can I get git to show me a list of untracked files whether or not they are ignored? 16:57
dukeleto Coke: git ls-files
tadzik dukeleto: I filed a GH issue about cardinal, also posted to parrot-dev 16:58
dukeleto tadzik: awesome! 16:59
17:01 plobsing joined 17:03 cogno joined
Coke looks like "make test-clean" isn't removing everything. 17:04
e.g. t/compilers/imcc/syn/pod_3.pbc
17:04 theory joined
Coke for those playing along at home, "git ls-files -o" is the incantation. 17:07
whiteknight: BuildTaskList is updated.
whiteknight Coke++ 17:08
dalek tracwiki: v19 | coke++ | BuildTaskList 17:11
tracwiki: remove stuff that's done, clean up what's left a bit.
tracwiki: trac.parrot.org/parrot/wiki/BuildTa...ction=diff
tracwiki: v184 | whiteknight++ | WikiStart 17:28
tracwiki: trac.parrot.org/parrot/wiki/WikiSta...ction=diff
17:48 M_o_C joined 17:54 chromatic joined 17:55 Kristaba joined 17:58 cogno left 18:03 lucian joined 18:04 bluescreen left 18:16 fbrito left 18:32 bluescreen joined 18:34 plobsing left
cotto_work ~~ 18:37
18:40 pmichaud joined 18:41 cogno joined
cotto_work seen fbrito 18:41
aloha fbrito was last seen in #parrot 25 mins 16 seconds ago leaving the channel.
dalek rrot: aeacfec | Whiteknight++ | t/dynoplibs/io.t:
fix some stat tests. On win32 we are looking for the file 'parrot.exe', not 'parrot' like on unix
 18:42
rrot: 5d54e34 | Whiteknight++ | t/dynoplibs/io.t:
comment out tests that cause premature process abort on win32.
18:43 darbelo joined
dalek rrot: ca72cd0 | Whiteknight++ | t/dynoplibs/io.t:
update test count
 18:44
18:51 mtk left, mtk joined
dalek TT #1933 created by whiteknight++: stat opcode causes program crash on Win32 with invalid file descriptor 18:51
TT #1933: trac.parrot.org/parrot/ticket/1933
19:05 bluescreen left 19:22 bluescreen joined, rfw joined 19:25 fbrito joined 19:27 jan left 19:30 cogno left 19:32 bacek joined 19:33 nwellnhof joined, jan joined
nwellnhof hey, whiteknight had another commit with only letters in the commit hash. 19:34
whiteknight awesome
nwellnhof i had one recently. plobsing computed the chances to 0.1%. 19:35
cotto_work he probably did it on purpose
19:36 plobsing joined
rfw morning cotto_work 19:39
which test was broken again?
19:41 mtk left
cotto_work rfw: just a sec 19:41
nopaste "cotto_work" at 192.168.1.3 pasted "for rfw" (56 lines) at nopaste.snit.ch/27554 19:42
cotto_work the test names should also be unique when possible
that makes it easier to figure out which test is broken quickly
rfw oh, haha
i
i almost saw that coming 19:43
19:45 cogno joined
rfw cotto_work: fixed 19:47
cotto_work rfw: all pass 19:49
rfw yay
cotto_work names look better too
I'll review shortly.
rfw alright 19:50
19:55 mtk joined
cotto_work rfw: did you test on windows? 19:55
rfw erm, let me get on that now 19:56
parrot is rather tempermental on windows
cotto_work indeed
I have a windows machine I can use. 19:57
rfw let me fire up cygwin
tadzik ~~ 19:58
19:59 TonyC left, nopaste left
rfw oh hey, parrot is actually compiling 20:02
cotto_work it tends to do that
rfw /usr//usrlib//libgcc//gcci686/-i686pc--pccygwin-/cygwin4.3.4//4.3.4./../../../../../../../.i686/-i686pc--pccygwin-/cygwinbin//binld/:ld :cannot cannotfind find- lparrot 20:03
so much for actually compiling
and now it's compiling again
0 [main] parrot 15460 exception::handle: Exception: STATUS_ACCESS_VIOLATION 20:04
764 [main] parrot 15460 open_stackdumpfile: Dumping stack trace to parrot.exe.stackdump
20:04 fbrito left
cotto_work rfw: gist.github.com/765320 20:04
rfw stupid non-conformist operating systems 20:05
cotto_work Yeah. That Linux thing is really annoying.
;}
rfw okay i think i'll just not check for output with filesystem erros 20:06
cotto_work that strikes me as a good idea
return code is fine
rfw the can_execute test is broken for reasons i don't know, though 20:07
20:07 plobsing left
whiteknight can_execute is broken? What platform? 20:07
rfw whiteknight: win32
so i'll skip it
whiteknight oh yeah, I am fixing tha tnow
rfw but that's odd, because parrot an--oh
whiteknight or, I was before I got distracted by socket error failures
actually, I think I already fixed the can_execute failure. github.com/parrot/parrot/commit/ae...a600656542 20:08
sockets are hellabroken on win32 20:09
rfw oh, the cwd test fails on win32 because you can't delete the directory you're executing in 20:10
whiteknight they may be broken everywhere, I think the tests are based on some bad assumptions
rfw i'll just skip that in win32
dalek rrot: 1c175e8 | Whiteknight++ | src/ (2 files):
we are testing whether a socket is opened or closed by comparing to a closed value, but we never initialize. Fix that
 20:13
whiteknight on win32 I think headerizer screws up all the whitespace
nwellnhof whiteknight: aaargh, your last commit has windows newlines 20:15
src/io/socket_api.c | 776 +++++++++++++++++++++++---------------------- 20:16
src/pmc/socket.pmc | 878 ++++++++++++++++++++++++++-------------------------
20:17 cogno left 20:19 tcurtis joined
whiteknight nwellnhof: yeah, I'm fixing that now 20:19
I thought I had git on that platform configured to avoid that
rfw hooray, it works
# got: '. .. sillynameā˜ŗ'
parrot just needed to be a little happier 20:20
cotto_work #ps in 9 20:22
dalek rrot: 581131f | Whiteknight++ | / (3 files):
make headerizer, and fix line-endings
whiteknight okay, when I initialize handle values in Socket, tests start failing as I expected 20:23
so we've been relying on broken behavior to make the tests look good for several months
rfw dear cygwin, please stop sucking so much 20:24
20:24 kid51 joined
rfw cotto_work: i can't get the io dynoplibs working because cygwin keeps STATUS_ACCESS_VIOLATIONing on me 20:26
cotto_work nice
Coke Do we have anyone building on cygwin regularly? 20:27
we need to resurrect taptinder. (mj41++)
nwellnhof coke: fperrad++ i think
Coke nwellnhof: for cygwin? k. 20:28
kid51 #parrotsketch in 2
Coke perhaps someone braver than I could poke rurban about it.
20:28 perlite_ joined
kid51 Any GCI students on channel? Feel free to join us in #parrotsketch 20:29
mikehh #ps time 20:31
dalek rrot: 7295dd1 | Whiteknight++ | / (2 files):
get_bool on a Socket returns whether the Socket is open. The is_closed method returns the opposite value. Tests for creating a new socket shouldn't use the get_bool vtable, since that shows whether the socket was opened, which these tests aren't doing
rfw cotto_work: all fixed on cygwin 20:32
20:32 perlite left
cotto_work rfw: great 20:32
20:32 perlite_ is now known as perlite
rfw cotto_work: pushed 20:32
cotto_work rfw: still explodes after test 8 with "The symlink function is unimplemented at t/dynpmc/os.t line 266." 20:34
rfw oh, you're using mingw32 aren't you 20:35
god i hate windows so much
cotto_work msvc/strawberry and mingw/strawberry 20:36
rurban why so shy 20:37
Do I bite that much? 20:39
rfw cotto_work: that should do it
cotto_work rurban: ? 20:40
rurban coke didnt dare to ask me something. what?
cotto_work rfw: bad plan (33 vs 34) but otherwise good
rfw cotto_work: my plan is good 20:41
rurban I'm just smoking my perl5 compiler, I cannot smoke parrot on cygwin today
cotto_work rfw: on msvc only 33 tests are fun
run 20:42
odd typo
rfw aha, got it 20:43
cotto_work: done 20:44
Coke rurban: you can be a little sharp, sure. I was mainly just dodging anything involving effort atm due to $DAYJOB
but it sounds like things are improved already. 20:45
rurban symlink not defined in os.t is very old, I remember
whiteknight does cygwin even support symlinking? 20:51
dalek rrot: d7eaaba | rfw++ | t/dynpmc/os.t:
[t] Fixing OS dynpmc tests on cygwin
rrot: de4d3e2 | rfw++ | t/dynpmc/os.t:
[t] Fixing OS dynpmc tests to work on other Windows silliness
rrot: 9f0a8b9 | rfw++ | t/dynpmc/os.t:
[t] fixing bad MSVC plan
rrot: e25181b | cotto++ | t/dynpmc/os.t:
Merge branch 'os_dynpmc_coverage_2' of github.com/rofflwaffls/parrot into rofflwaffls-os_dynpmc_coverage_2
rrot: a7c3149 | cotto++ | t/dynpmc/os.t:
Merge branch 'rofflwaffls-os_dynpmc_coverage_2'
whiteknight t/dynoplibs/debug.t and t/tools/pbc_disassemble.t are both failing tests on win32. 20:52
cotto_work rfw: approved
rfw thanks 20:53
whiteknight socket tests are all passing on win32, for the first time in months
dalek rrot: 7b94fae | Coke++ | src/string/encoding/shared.c:
remove tabs.
rrot: c5013cf | cotto++ | t/dyn (2 files):
[t] codingstd fixes in recent tests
 20:59
21:01 plobsing joined 21:02 bacek_ joined 21:06 bluescreen left
Coke thinks whiteknight should run "make codetest" more often. ;) 21:07
21:07 bacek left
whiteknight It's easier on my normal dev machine, where all the text editors are configured with the right settings 21:08
cotto_work mikehh: codingstds needs some love if you have the tuits
whiteknight I've been bouncing back and forth between environments and half-configured editors
mikehh cotto_work: have a look in a minute 21:09
cotto_work mikehh: thanks
21:09 jsut joined 21:11 fperrad left 21:14 jsut_ left 21:22 TonyC joined 21:24 nopaste joined
Kristaba Someone claimed the fill_params task? 21:26
21:26 whiteknight left
cotto_work Kristaba: no. I just unpublished it. 21:26
Kristaba cotto_work: Oh, you think it's not a good GCI task or it's for another reason? 21:28
dukeleto ~~
cotto_work It's not a good gci task.
21:29 bacek__ joined
cotto_work Changes to that code require intense scrutiny and care. 21:30
21:30 Yuki`N joined
dalek rrot: 4e20ec0 | jkeenan++ | src/io/socket_api.c:
[codingstd] Needed closing paren in POD.
 21:30
chromatic I think you're all making too much of that code. It's super easy to break tests there.
dukeleto chromatic: good to see you around 21:31
cotto_work +1
Kristaba Yes, I understand, no problem
(It would have been interesting and fun, but not necessarily a good thing for Parrot :p) 21:32
cotto_work Kristaba: if you want to take a shot at it, we can make it a retroactive task. I'm still going back and forth on whether it'd actually be good for a gci student. 21:33
Yuki`N So what exactly should be printed to config.log. 21:35
Everythinng?
*everything including the header stuff?
21:35 bacek_ left
cotto_work Yuki`N: yes. There's no reason to be sparse when spitting to a file since we can assume that users have grep or a decent text editor. 21:36
Kristaba cotto: Thank you for the proposal, if I'll want to do this I'll tell you. 21:38
21:38 rurban_ joined
cotto_work Kristaba: ok 21:39
Yuki`N I wish Perl was like Python and you could redefine the print function.
I'm not sure sure how I'm going to do this anymore..hm/
21:41 rurban left, rurban_ is now known as rurban
mikehh opbots, names 21:43
21:45 Andy joined
dukeleto Yuki`N: sub print { ... } 21:45
Yuki`N: the perl 5 debugger redefines the sub named 'sub' ;) 21:46
dukeleto has read the source of the perl 5 debugger, fixed a bug in it, and lived to tell the tale
Yuki`N dukeleto, aha.
dukeleto we are almost at pull request 100. Wow. 21:47
Yuki`N Where are the actual config steps declared?
Coke lib/Parrot/Configure/Step/List.pm 21:48
Yuki`N Where's the code that's run, though?>
Coke you said declared. :)
21:49 tcurtis left
Coke they are defined in config/<name1>/<name2>.p 21:49
they are defined in config/<name1>/<name2>.pm
dukeleto bacek_at_work++ tadzik++ # cardinal hacking
Yuki`N: the config steps are a deeeeeep rabbit hole. Have fun :)
tadzik is excited about being on #ps :)
Yuki`N Oh dear. 21:50
Coke eh. it's not super crazy.
dukeleto realizes he has missed #ps
Coke ask and we'll answer.
dukeleto backlogs
Yuki`N I think redefining print will work for my needs, :P 21:52
21:53 Patterner left, Psyche^ joined, Psyche^ is now known as Patterner
Yuki`N Although, there's no way for me to see whether or not verbosity is set... 21:54
21:55 kid51 left, darbelo left
Yuki`N Oh wait, there's this cool filter package. 21:56
Oh, it's in CPAN, hmm. 21:58
21:59 M_o_C left
dukeleto pmichaud: i am working on the "test HLLs" issue (backlogging from #ps today) 22:01
pmichaud: i plan on using rakudo's "make test" and nqp-rx's "make test" on parrot master as the first tangible goals 22:03
Coke any chance we can revert the recent :main changes (or whatever introduced the borkage?) 22:04
and then only re-apply them once the testing bears it out?
dalek rrot: df64b9a | mikehh++ | src/io/socket_api.c:
fix codetest failure - pod syntax and c_func_docs
dukeleto pmichaud: i could have this done in a few hours if i had uninterrupted time to work on it. So it may take a few days
do y'all want me to revert the encapsulate-main merge ? 22:06
someone told me it was reverted, but I am not seeing it that way 22:07
the encapsulate-main merge was cbaad3757d3d5b5c21f5a334d4372801e5cde879
dukeleto hears crickets 22:09
dalek rrot: d69e236 | mikehh++ | t/dynpmc/os.t:
fix perlcritic failure, replace two argument open(s) with three argument open
 22:11
Coke I thought someone anti-cherry-picked something post merge. 22:12
(but that the mergeback as a whole was untouched)
Yuki`N Do you want --verbose=2 going to config.log? 22:13
Coke I would imagine any amount of verbosity would just go to the log. 22:15
but as was mentioned on the ticket, I have little imagination on this ticket. 22:16
dukeleto Yuki`N: yes, as much info as possible, for now 22:17
Yuki`N: we can ratched down the verbosity later if we want 22:18
Coke: i don't see any anti-cherry-picking about encapsulate-main
cotto_work: encapsulate-main . Does it need to be reverted ?
Yuki`N I'm still not quite sure how I can pull this off.
cotto_work dukeleto: it was, partially 22:19
Yuki`N There's really no way as it holds to mark some prints verbose and some prints normal.
dukeleto cotto_work: in what commit?
22:19 bacek_ joined
plobsing I didn't revert it to get implicit-:main back. I augmented it. 22:19
dukeleto cotto_work: i don't see evidence of the partial revert, but i could be blind
plobsing searches for commit
Coke "it was, partially" - that's what I said. er, meant.
plobsing fdb6fa57549b91ab6024d046909fff7eedfdb75d 22:20
Yuki`N Just the way verbose printing is handled in config makes this impossible without rewriting every print line. :/ 22:21
dukeleto Yuki`N: don't worry about it. Just make a config.log that useful in some way :) 22:22
Yuki`N I'd probably end up re-doing the print system. 22:23
22:24 bacek__ left
dukeleto Yuki`N: don't go down that road 22:25
Yuki`N: just generate the simplest-yet-useful config.log for this task
Yuki`N The current architecture makes it impossible.
dukeleto Yuki`N: we want small useful iterations that can be improved in the future
22:25 Matt221 joined
dukeleto Yuki`N: why can't you just duplicate all output from Configure.pl and put it into config.log ? 22:26
Yuki`N: create a dup of STDOUT/STDERR and redirect to config.log . Done.
Yuki`N Well that would be the same as running `perl Configure.pl | tee config.log`
cotto_work why is that bad? 22:27
Yuki`N Wasn't the point to have verbose info feed to config.log, while terse info goes to console unless --verbose is specified?
dukeleto Yuki`N: basically. but we don't want users to have to redirect the output themselves. We want it to be "automagical"
Yuki`N: that would be nice, but if that is hard, let's just do the simple option first
dukeleto migrates back home from the coffee shop 22:28
Yuki`N Oh, ok. 22:30
That's actually easy then.
dukeleto yay, easy!
dukeleto actually goes
Yuki`N There's a CPAN module that does that...hmm.
I really want to use it, but >dependencies.
dukeleto Yuki`N: be careful about new dependencies
Yuki`N: we can include CPAN modules in the parrot repo, but we need to think carefully about that
Yuki`N The concept is simple, the source is 3kb w/ documentation.
dukeleto actually actually goes
Yuki`N I will discuss later then. 22:31
22:35 simon_ left
Kristaba cotto: Finally, I claimed the zero-arity task, and I have already a question... 22:37
cotto: There are a lot of test fail, as expected, but really I don't understand why the imcpasm tests fail that way 22:38
22:38 nwellnhof left
Kristaba cotto: What am I supposed to do with that? Stupidly copy the many "get_params" into the expected output? 22:38
cotto_work Kristaba: building now so I can play along at home 22:39
ok. which test now? 22:42
22:42 Matt221 left 22:43 Matt221 joined
Kristaba t/compilers/imcc/imcpasm/opt1.t for example 22:43
dalek TT #1934 created by mikehh++: Some Python programs added to tools/dev 22:45
TT #1934: trac.parrot.org/parrot/ticket/1934
22:46 davidfetter joined
cotto_work bacek_: ping 22:47
bacek_ cotto_work, pong 22:48
cotto_work bacek_: Kristaba has some questions about zero-arity test fixes 22:49
bacek_ ah, did you see my msg about "difficult task"? :)
Kristaba, opt1/opt2 in imcc? 22:50
cotto_work yes. I changed the difficulty
bacek_ Kristaba, it's about testing optimizer in IMCC. Which looks like broken.
cotto_work urgh 22:51
Kristaba bacek_: Yes, indeed :(
bacek_ Kristaba, ignore them for now. There are many other tests which can and should be fixed.
Kristaba bacek_: Ok, no problem 22:52
bacek_ Kristaba, mostly related to #1705 and #1033
cotto_work bacek_: should those failing tests just be TODO'd? 22:53
or skipped
bacek_ cotto, not yet.
Best way - implement PIRATE and strip IMCC. Or at least strip it down to bare minimum for bootstrapping.
pmichaud dukeleto: (from backlog) TT #1704 has the history of where we stand. (more)
the branch merge was commit cbaad37 22:54
22:54 plobsing left
pmichaud in order to "fix" the problems introduced by the merge, plobsing++ re-added the implicit :main capability in fdb6fa5 22:54
that appears to have gotten rid of the "No main sub found" problem, but now we have difficulties with mainline code being executed twice 22:55
it only appears to affect p6/nqp programs that define new classes 22:56
since p6 always declares/defines new classes, that's just about all of them :-)
nqp's tests only have a few tests that declare new classes -- those are the tests in which we end up with double-execution of main
the thing that is significant about classes defined in nqp is that they're established by :load/:init subs, if that makes a difference. 22:57
(whereas other nqp programs that don't declare :load/:init subs don't appear to have the double-main execution)
I'm not sure if they're related, but there appears to be a correlation there.
(end)
22:58 Matt221 left
pmichaud as far as whether we need to completely revert the branch merge, or patch things to work going forward... I don't know the pain points involved with either so I think that's really up to others to decide. 22:58
also, thinking about it a bit more.... given that the tests that failed in nqp are the ones that declare new classes, I suspect the problem isn't really the lack of implicit :main (although that will likely cause problems for rakudo modules and standalone-compiled code) 23:00
actually, disregard that last statement
since nqp-rx was seeing the "No main sub found" errors in its tests, that still points to a problem with :main somewhere
cotto_work bacek_: what's the best thing for Kristaba to do about those tests? 23:07
23:08 bacek_ left 23:09 hercynium left
cotto_work Kristaba: ignore them for now. I won't let them block your task completion if we can't give you a good answer wrt what to do with them. 23:09
mikehh All tests PASS (pre/post-config, make corevm/make coretest, smoke (#2023) fulltest) at 2_11_0-683-gd69e236 - Ubuntu 10.10 i386 (gcc-4.5) 23:10
23:12 lucian left
Kristaba cotto: Okay, no problem I'll try to fix the other test for now 23:14
23:16 dmalcolm left 23:20 bacek_ joined
Kristaba bacek_: Sorry to bother you again, but I think there is an other "problem" 23:27
bacek_ Kristaba, which one?
Kristaba The test at at t/compilers/imcc/syn/pcc.t line 525 is "calling a sub with way too many args", and as its name suggests, it try to call a subroutine (with 0 parameters) with 2000 arguments, and it expects to succeed oO
I think it's the exact opposite of the goal of this branch, isn't it? 23:29
bacek_ Kristaba, yes. Test is correct. It test absence of segfault.
It should say "Too many args" (or something similar) 23:30
hang on
I probably on wrong branch
No, still correct answer. Test check for segfaults 23:31
But should be rewritten to properly test it. 23:32
Kristaba bacek_: But now, when a function with 0 parameter is called with arguments, it returns an error :/
bacek_ Kristaba, yes. "should be rewritten" :) 23:33
cotto_work I remember that test.
Kristaba Ok, so I ignore it too? 23:34
Or I rewrite it?
bacek_ E.g. push_eh bang; foo(_ARGS_); pop_eh; bang: say "didn't segfault"
cotto_work Kristaba: just check that it doesn't segfault.
bacek_ Or something like this
Kristaba, nope, fix the test :)
Kristaba Okay :p 23:35
bacek_ cotto, ping 23:42
cotto_work bacek_: pong 23:43
bacek_ cotto, speaking of opmap_aware_pmc. What is relation between PackfileOpMap and OpLib? 23:44
(In your design)
cotto_work bacek_: do you understand how the dynop mapping code works? 23:45
bacek_ barely
Looks like PFOM "has many" OpLib. Correct?
cotto_work let me refresh myself
bacek_ And for each Op we have tuple (oplib, index) as mapping 23:46
cotto_work yes and yes
23:47 plobsing joined
cotto_work each set of op maps has a single OpLib 23:47
bacek_ hmmm 23:48
Ok. Let me dig little bit deeper
cotto_work alright 23:49
23:59 gbacon left

Powered by Cro and the RakuĀ® Programming Language.

Please report any issues / comments / feature requests as an issue on App::Raku::Log.

Thank you!